Jump to content
chiliwili69

Valve Index vs. HP Reverb (through the lens pictures)

Recommended Posts

There is already a nice thread about this same topic but I want to create a new one just for the pictures I obtained and other test that I will do, similar to the comparison I was doing to the Pimax5K+ and Index in this other thread

 

I reused the same recorded tracks (they still work in the latest IL-2 version) to obtain the through the lens pictures with the same compact Panasonic Lumix camera in the Reverb. And I used the previous pictures of the Index from the Pimax5K comparison.

 

All raw images can be downloaded from here: https://drive.google.com/open?id=1p6YMODm_oQBITJRuJVx11iKwOHvzG-6r

 

I set the Reverb SteamVR SuperSampling to 188% which is a bit above to what my graphic card (1080Ti) can deliver, but I just wanted to obtain the best graphics quality (perhaps I could increase it even more, but I will analyze the SS effect in the Reverb in future tests). 

In the Index I use 140% which is 12.6 million pixels and for the Reverb I use 188% which is 9.5 million pixels. Surprisingly the Index gets a little bit better performance, but I will analyze this in future tests. This table compare the different rendered pixels for every SS ratio (1.88 means 188%).

resolution.png.9dcade7d0b34b0e80320b11228486653.png

 

IDENTIFICATION:

01-Compo-ID.thumb.png.b1c0a2a2b42e232a0f9a43d2b6aabddf.png

 

GROUND SPOT and DETAILS:

02-Compo-Ground.thumb.png.3fbf02b021d87b27d75ba23a3f1b7958.png

 

Edited by chiliwili69
  • Thanks 2

Share this post


Link to post
Share on other sites
10 minutes ago, chiliwili69 said:

 

GAUGES:

03-Compo-Gauges.thumb.png.346893ef6b9ad2ccab29adab10a8e03e.png

 

DETAILS PLANE:

04-Compo-Plane.thumb.png.8df0a2ac62c7a8d094dc315fc76889e8.png

 

 

DETAILS SHIP:

05-Compo-ship.thumb.png.f94efec5fdadda46642740182ec2e2bc.png

 

Thankyou for this! Just 100% sold me on the reverb. 

Share this post


Link to post
Share on other sites

As you can see the SDE is practically eliminated in the Reverb. In the Index you can still see it, but it doesn´t affects me a lot.

Looking only at pictures one will quickly choose the Reverb due to the increased level of detail, but there are other important aspects to consider: FOV, performance, comfort, sweet spot, glare, tracking, audio

Yesterday I was playing about 1 hour with each device and it is still very difficult to choose one.

 

One thing that clearly arise was the weight of the device, the Reverb is really light and this helps a lot for long flying sessions.

 

In any case, it is great that VR has reached a level where we have good resolution, good FOV, good confort, good audio, etc, etc. I am sure in a couple of years (or less) we will have a device that compile all of this.

It is a pitty that with the good lenses of the Index and FOV, they "only" went to the VivePro or O+ resolution. But it was a compromise since for most of other non-flight sim games it could be enough.

 

Share this post


Link to post
Share on other sites

Great images.

 

Though I think the 188% vs 140% is always going to confuse.

 

188% using the non WMR beta and the Reverb equates to a resolution of around 2200x2160 per eye (hence 9.5 million)

140% using the Index equates to a resolution of 2384x2648 per eye (hence 12.6 million)

 

The scaler is off in SteamVR for the Reverb and fixed with the beta version of WMR for SteamVR (where ~74%) yields the same resolution

Share this post


Link to post
Share on other sites

Maybe better lenses which improve the “sweet spot” of the reverb will appear as a mod in the future? :) 

Share this post


Link to post
Share on other sites

Well, hard to argue with those images. If I were going to choose, the Reverb would win on combination of image clarity and lack of base station requirement. I am using VR exclusively for BoX. I don't see a need to buy now however. Maybe if the Reverb were on sale I might give it a try. Newest offerings still look to be at gen 1.3 to 1.5. Lack of mechanical IPD adjustment is just insulting. It cannot be justified as a responsible practice and the range of adjustments should be increasing not diminishing.

Edited by Dagwoodyt

Share this post


Link to post
Share on other sites
16 hours ago, Dagwoodyt said:

Well, hard to argue with those images. If I were going to choose, the Reverb would win on combination of image clarity and lack of base station requirement. I am using VR exclusively for BoX. I don't see a need to buy now however. Maybe if the Reverb were on sale I might give it a try. Newest offerings still look to be at gen 1.3 to 1.5. Lack of mechanical IPD adjustment is just insulting. It cannot be justified as a responsible practice and the range of adjustments should be increasing not diminishing.

There's a new headset coming out with the same screens and lenses but with adjustable IPD, can't remember who's making it. Think it's ACER. Been kind of waiting for that to come out. The pictures tell it all the reverb looks incredible. People were not exaggerating when they said it was nearly 1080p monitor quality. That first picture with the pe-2's it's unreal how good it looks. Feel sorry for people who "upgraded" to the rift s.

 

The big question is hows the damn performance. We have the screens now, but how do these things run. I can't imagine they run great considering the pixel counts. Also big thanks to chiliwili for always coming through with the cold hard facts and cutting through all the bullshit.

Edited by JonRedcorn

Share this post


Link to post
Share on other sites

2.jpg.d4ea3b8b453a4a88cac92afa301729e1.jpg

Pasted from a previous thread chili started. For clarity comparison, how that same ship looks on a 4k screen, 2x aa.

 

53 minutes ago, JonRedcorn said:

There's a new headset coming out with the same screens and lenses but with adjustable IPD, can't remember who's making it. Think it's ACER.


Acer OJO I think.
 

Share this post


Link to post
Share on other sites

Ignore the SS settings listed and look at Chili's total pixel count, the Reverb is doing more with near a third less rendered pixels, Index 12.6 mpixels vs Reverb 9.5 mpixels.  Most likely comes down to screen quality and higher native resolution panels of the Reverb compared to the Index.

 

I would expect a similar outcome in the near future with the Pimax 8K vs the soon to be announced Pimax 8K-X.  The better the screens and native resolution, the less you have to push SS settings to deliver a decent image.👍

Share this post


Link to post
Share on other sites

One thing I wanted to measure was the weight of both devices. 

These are the facts:

Index: 792 grams

Reverb: 540 grams

322151008_Index-Reverbweight.thumb.jpg.7baf5efbf862bac8f5d19bf237df9928.jpg

 

So the Index is about 250 grams heavier than the Index and this is something you notice well when trying both one after the other.

This factor affects the comfort. It is a pity that the extra things of Index (IPD mechanical, push/pull lenses adjust) has the extra cost of the weight. Nothing is ideal and there are subtle differences when wearing the devices.

 

Regarding comfort, both are good in general terms, but not fully perfect.

 

INDEX: After the IPD adjust (which works very well) you can adjust the distance of lenses to your eyes, but normally you always try to put it as close as possible to have more FOV. This mechanism works very well also. Then you need to fix the headset with the rear wheel. I always try to secure it very well to avoid any kind of lash/play between my head and the headset when I do quick turns of my head. But the Index, due to its "heavy" weight requires to really press it very against my face, the result is that after one hour of flying I start to notice some fatigue over my face so I have to decrease the pressure of the back wheel, but then I feel like it is not totally fixed to my head. I would say that the weight of the Index is not as well compensated as in the Rift CV1 or the HP Reverb.

 

REVERB: Apart from the static mass itself, the rotational inertia mass for quick turns is logically less, so you need to put less pressure over the face to keep the headset fixed. I find easier to reach the perfect adjust in the Reverb (just 5-10 seconds in Reverb, few more seconds in the Index) since it uses Rift CV1 style fixing, which I consider the best fixing mechanism.

But the Reverb is neither perfect, in my opinion it has two little defects depending on everyone:

- The lenses are very near to the nose and you need to set the upper velcro well fixed to avoid to touch the nose.

- The face cushion is a bit thin (in Reverb Pro). I really would like to have something thicker for long flying sessions since after one hour I notice the pressure over my eyebrows.

 

The worst thing is I have absolutely no idea which one will I keep. Let´s log more hours and do more tests.

Share this post


Link to post
Share on other sites

Chiliwili,

 

When using the Reverb for flight sims (actually in the cockpit flying), are the details noticeably clearer?  Because in those images it appears substantially better. 

 

I almost bought a Reverb but the smaller FOV (and the teething issues) kept me away.  

Edited by WIS-Redcoat

Share this post


Link to post
Share on other sites
On 9/2/2019 at 11:08 AM, SvAF/F16_radek said:

Chili what's your verdict on godrays in the reverb? (vs index)

 

There are also some godrays (glare) in the Reverb, maybe a bit less present that in the Index but in the IL-2 menu (black and white) there is also some galre with the REverb, so they are not like the Pimax where the glare was almost zero.

In any case, I have tried also night scenarios with moon, no moon, sunrise, etc and the effect in any device (Reverb or Index) doesn´t affect me a lot since it is rarely noticeable.

On 9/2/2019 at 3:37 PM, blitze said:

Ignore the SS settings listed and look at Chili's total pixel count, the Reverb is doing more with near a third less rendered pixels, Index 12.6 mpixels vs Reverb 9.5 mpixels.  Most likely comes down to screen quality and higher native resolution panels of the Reverb compared to the Index.

 

Yes, you are right. If the panel has a good resolution (like Reverb 2160x2160) you really don´t need to supersample the image, you can just render 2160x2160 per eye and it is enough (paradoxically in the steamVR you need to set it to 188%).

This is the same story than a 4K monitor, the 4K is so good that you don´t need to supersample or to introduce big AA filter.

3 hours ago, WIS-Redcoat said:

When using the Reverb for flight sims (actually in the cockpit flying), are the details noticeably clearer?

 

Yes and No.

Yes for the center of the view, but No for the area around the center.

I have been comparing the raw images captured by the camera (the link is above) and they don´t show the sweet spot effect or chromatic aberration in edges of the Reverb.

The Reverb images captured by the camera are clearer in the edges, but this doesn´t correspond with the reality.

 

 

 

Edited by chiliwili69
  • Like 1
  • Upvote 1

Share this post


Link to post
Share on other sites

Well, I am starting to have a preference now.

This afternoon I have been playing with the Index about 1 hour, same QMB in spring Kuban sea at sunrise (and also night), doing exactly the same all time: with the Spitfire killing 110s, the train and ships. 

Just after that, I switched to the Reverb and I quickly noticed several worse things in the Reverb:

- FOV: The vertical FOV of Reverb is much much smaller than Index, and also the horizontal. I really liked the FOV of the Index and I was missing that after the switch. Larger FOV is really useful.

- Sweet spot: I noticed that very quickly without looking for it. In the Reverb the images gets blurred and chromatic aberration as your eyes move from enter of view. Period.

- Performance: I still have not run the tests, but with 140%SS at Index at 80Hz (12.5 million pixels), the performance is much better in the Index than in the REverb with 188%SS at 90Hz (9.5 million pixels). I know that running at 80 helps to the CPU and GPU, but it really doesn´t explain why the performance is so much worse in the Reverb. So performance wise it is better the Index

- Foam cushion: The synthetic leather  doesn´t allow to breath the skin, and if your skin is hot you will get some fogging on the lenses.

 

But the curious thing is that after 15 minutes playing with the REverb, the smaller FOV, worse sweet spot, and even worse performance was less a less annoying. So I continued playing for another 45 minutes and enjoyed the great view on the center. Then I changed the SS to 140% to have better performance, but the image was a bit less sharp and clear.

 

Then I switched again to the Index for 30 minutes and in the initial moments I quickly realize that THE INDEX IS STARTING TO BE MY FAVOURITE HEADSET!!!

 

Why?

 

1.(40%) bigger FOV is GREAT (I already lost a FOV portion going from Pimax to Index, but I am not ready to lose more FOV).

2.(40%) bigger Sweet spot is GREAT. This together with the bigger FOV gives you a better feeling of presence in the scene.

3.(20%) Better comfort. I have been learning to to adjust the top strap with the right pressure in the back wheel. So, I feel better with Index.

 

The % indicate the weight of the reason. I didn´t include performance since this could be improved by acquiring a 9700K and a 2080Ti.

 

It is not a definitive choice, but just a preference for now.

Edited by chiliwili69
  • Thanks 1

Share this post


Link to post
Share on other sites
3 hours ago, chiliwili69 said:

Well, I am starting to have a preference now.

This afternoon I have been playing with the Index about 1 hour, same QMB in spring Kuban sea at sunrise (and also night), doing exactly the same all time: with the Spitfire killing 110s, the train and ships. 

Just after that, I switched to the Reverb and I quickly noticed several worse things in the Reverb:

- FOV: The vertical FOV of Reverb is much much smaller than Index, and also the horizontal. I really liked the FOV of the Index and I was missing that after the switch. Larger FOV is really useful.

- Sweet spot: I noticed that very quickly without looking for it. In the Reverb the images gets blurred and chromatic aberration as your eyes move from enter of view. Period.

- Performance: I still have not run the tests, but with 140%SS at Index at 80Hz (12.5 million pixels), the performance is much better in the Index than in the REverb with 188%SS at 90Hz (9.5 million pixels). I know that running at 80 helps to the CPU and GPU, but it really doesn´t explain why the performance is so much worse in the Reverb. So performance wise it is better the Index

- Foam cushion: The synthetic leather  doesn´t allow to breath the skin, and if your skin is hot you will get some fogging on the lenses.

 

But the curious thing is that after 15 minutes playing with the REverb, the smaller FOV, worse sweet spot, and even worse performance was less a less annoying. So I continued playing for another 45 minutes and enjoyed the great view on the center. Then I changed the SS to 140% to have better performance, but the image was a bit less sharp and clear.

 

Then I switched again to the Index for 30 minutes and in the initial moments I quickly realize that THE INDEX IS STARTING TO BE MY FAVOURITE HEADSET!!!

 

Why?

 

1.(40%) bigger FOV is GREAT (I already lost a FOV portion going from Pimax to Index, but I am not ready to lose more FOV).

2.(40%) bigger Sweet spot is GREAT. This together with the bigger FOV gives you a better feeling of presence in the scene.

3.(20%) Better comfort. I have been learning to to adjust the top strap with the right pressure in the back wheel. So, I feel better with Index.

 

The % indicate the weight of the reason. I didn´t include performance since this could be improved by acquiring a 9700K and a 2080Ti.

 

It is not a definitive choice, but just a preference for now.

Another concern I have is tracking. Over last few days my Odyssey is starting to have tracking issues. My hope is that it is related to a recent Win10 update and will eventually resolve itself. Frequently I am having to disconnect the USB cable and reinsert it before the Odyssey is recognized. I have had to revert to the O+ which seems not to  have the tracking issue for now. I am imagining a situation in which the Reverb might also become camera glitchy and thinking how that might be worsened given its' smaller fov. I am wondering if it is possible to get by using an Index with one 2.0 base station placed above my monitor. The Index larger field of view sounds attractive.

 

Share this post


Link to post
Share on other sites

Is the 2080 (non-ti) enough to drive the Index for IL2? I am still on Rift S, fairly happy with it, but the increased FOV of Index is tempting.

Share this post


Link to post
Share on other sites

The leather face band is why I’m waiting for the consumer edition. :salute:

Share this post


Link to post
Share on other sites
6 hours ago, Dagwoodyt said:

Another concern I have is tracking. Over last few days my Odyssey is starting to have tracking issues. My hope is that it is related to a recent Win10 update and will eventually resolve itself. Frequently I am having to disconnect the USB cable and reinsert it before the Odyssey is recognized. I have had to revert to the O+ which seems not to  have the tracking issue for now. I am imagining a situation in which the Reverb might also become camera glitchy and thinking how that might be worsened given its' smaller fov. I am wondering if it is possible to get by using an Index with one 2.0 base station placed above my monitor. The Index larger field of view sounds attractive.

 

 

The tracking with the Reverb hasn't missed a beat, nor has the Index with a single 1.0 base station*! I set mine high and up to the left. I am surprised with both how far I can turn my head around without losing tracking.

 

*As long as I don't touch it, I get a lot of EMI issues with the Index where it will go grey if I touch the headset, though I do have a lot of kit including a direct drive steering wheel which is the likely cause. I have tried ferrite cores to no avail. No such issues with the Reverb.

Share this post


Link to post
Share on other sites
5 hours ago, Alonzo said:

Is the 2080 (non-ti) enough to drive the Index for IL2? I am still on Rift S, fairly happy with it, but the increased FOV of Index is tempting.

Not sure about Index users but my RTX2080 is driving approx 28.2Mpixel with FFR (Fixed Foreated Rendering) on my Pimax 8K.

At a guess FFR would probably halve the actual pixel count, it seems to be a 3 zone render approach (guessing 16.5Mpixels using center 3rd being full res, next step down being half that and the outer third being half that again).

 

Without FFR it renders 10Mpixels quite well but I also have a limitation on my CPU which is mid range and the bottle neck in Il2.

Share this post


Link to post
Share on other sites
15 hours ago, Alonzo said:

Is the 2080 (non-ti) enough to drive the Index for IL2? I am still on Rift S, fairly happy with it, but the increased FOV of Index is tempting

 

My 1080Ti do a decent work with the Index as you can check in the previous posts when i was comparing it to Pimax5K+.

 

A 2080 is equivalent to a 1080Ti, so i don´t think you will have problems (given also your CPU). But for sure the Index will demand more resources than the Rift-S. If you have currently some margin then you will be fine.

  • Like 1

Share this post


Link to post
Share on other sites

@Chilliwilli,

      As you have stated before, the snapshot through-the-lens picture for the Gauges case of the Reverb (Photo-2-Reverb-Gauges.JPG) does not seem to show any real blurriness or chromatic aberration at the edges (maybe even that snapshot is a bit cropped ). I wonder what that implies - our human eyeball lenses (maybe different from person to person?? ) don't act like a camera. If so, why would one HMD be different than another if the image on the screen is mostly perfect. Could it be the quality of the fresnel lens? Maybe it matters how good our eyesight is and whether we have a good glasses prescription or not (if we wear them).

 Also, how about the through-the-lens camera - does that have an effect?

 

Thanks much for the benchmarks. I am repeating them for comparison. So far, I get your results or maybe a little better - my PC is a 9700Kat 5Ghz +  RTX2070. My HMD at present is the O+.

Edited by whitav8

Share this post


Link to post
Share on other sites
On ‎9‎/‎4‎/‎2019 at 1:34 AM, chiliwili69 said:

THE INDEX IS STARTING TO BE MY FAVOURITE HEADSET

Your comparison really makes me willing to try the Index once I get a chance since after spending some hours in the Reverb I really can't see myself going for something with less visual clarity.

Apparently personal preferences do matter :)

 

In my view, the Reverb is by far not perfect. You name it - FOV, visible display edges, sweet spot... I am aware of those too but only once I put the headset on. The first minute into the flight the issues disappear :good: and the resolution and almost no SDE start shining.

 

In any case… good you can compare the two headsets directly. I am sure you can't make a wrong choice between the two and I hope the one you pick will serve you well! :salute:

 

Thanks for the comparison.

Edited by apollon01

Share this post


Link to post
Share on other sites

I cannot enjoy the VR experience if the sweet spot doesn't cover all the lens.
I suppose the next Gen VR will be finally a VR 2.0. And the headset should be lighter for sure. 

Edited by Fran13

Share this post


Link to post
Share on other sites

Many of us dream of better (and still affordable VR). The question is what inbetween?

 

The Oculus Rift kickstarted in 2012 only to bring the VR 1.0 to the market in 2016.

 

Since then (3 years) there were improvements on the way but not really groundbraking VR 2.0 as you say 😃

 

As it still may take some time, why not to have fun with VR 1.5? 😎

Share this post


Link to post
Share on other sites

Because is my money and I don't wanna be invest in a 1/2 improvement 😎 

 

 

I will keep the Oculus 1 for the moment thanks.

 

  • Like 1

Share this post


Link to post
Share on other sites
4 hours ago, whitav8 said:

As you have stated before, the snapshot through-the-lens picture for the Gauges case of the Reverb (Photo-2-Reverb-Gauges.JPG) does not seem to should any real blurriness or chromatic aberration at the edges (maybe even that snapshot is a bit cropped ). I wonder what that implies - our human eyeball lenses (maybe different from person to person?? ) don't act like a camera. If so, why would one HMD be different than another if the image on the screen is mostly perfect. Could it be the quality of the fresnel lens? Maybe it matters how good our eyesight is and whether we have a good glasses prescription or not (if we wear them).

 Also, how about the through-the-lens camera - does that have an effect?

 

The raw images I put it in the link above are RAW images (in jpeg format) provided directly by the SIM card of the camera, I didn´t cropped them. Also the camera didn´t capture all the FOV or edges of vision.

 

Faces are different from person to person, but this will not affect to the sweetspot/chromatic effect since I have put my one single eye in all posible spatial position (up/down, left/right, in/out) and the sweet spot effect is not corrected. So, it is not a "face" problem.

 

Eyeballs are different from person to person, so this could explain that some people see more or less effect. People with eyeballs functioning like my camera will not see blurriness in the edges. I remember that when doing through lens picture with Rift, the camera showed chromatic aberration in the edges that I was not seeing with my eyes in the headset (so with the Rift my eyes were better than the camera).

 

I don´t use glasses for more than 50cm objects and I think I have a good vision, perhaps not 20/20 but close.

 

All lenses bend differently different waves lengths, that´s the origin of chromatic aberration and blurriness. But VR devices take into account each color independently (with a profile for each RGB color) when they apply the lens distortion profile to the image displayed in the HMD display. Perhaps HP can provide different distortion profiles for different types of eyeballs people and correct this effect via software. But I really don´t know if the are really conscious about this problem (VivePro was also awful in the sweet spot).

 

In any case, The REverb in his current state shows that problem to my eyes. Unfortunately my eyes doesn´t work as my camera.

4 hours ago, whitav8 said:

Thanks much for the benchmarks. I am repeating them for comparison. So far, I get your results or maybe a little better - my PC is a 9700Kat 5Ghz +  RTX2070. My HMD at present is the O+

 

That´s why I thinbk it would be good to have a benchmark, to compare performance of different headsets and check if our PC is working as expected, and also to learn what it matter when upgrading PC, or to help adjust the settings properly.

Unfortunately it is difficult to produce a benchmark track that work always as in the real game. Devs could do that and put it in the game, so everybody could run it and compare. Or at least produce a fixed flight record produced by them and ready to be used in every new release of the game.

I did a poll about a year ago and 83% of people were in favour of having something:

https://forum.il2sturmovik.com/topic/39713-il-2-in-game-benchmark/

 

Share this post


Link to post
Share on other sites

@ChilliWIlli,

      This is a plot of SteamVR reprojection (auto, not motionvector) on/off with my 9700K@5Ghz+RTX2070+Odyssey+   Thanks again for your detailed analysis! Seems like SteamVR reprojection should be more responsive to returning to 90Hz.

1342545815_Bench1-Reprojection.thumb.jpg.cdbd93cfa7ebf5eda2746ffcf6e7967a.jpg

Edited by whitav8

Share this post


Link to post
Share on other sites

Well, as in the Pimax5K vs Index thread, It is now the time to analyze performance for IL-2 and my system with both devices, Index and Reverb.

My CPU is a 4790K running at 4.8GhZ with 1.3 Vcore. Fastest DDR3 RAM and 1080Ti. I could upgrade my system tomorrow if I wanted, but I still don´t see the upgrade will give me a significant enhancement for IL-2 VR. I am waiting to next gen CPUs and next gen GPUs for now.

 

I will use the same 2 tracks benchmarks used in the Pimax5K comparison. You can download all tracks from here.

BENCH1: 110 seconds, Spitfire in Kuban summer Novorossiysk against some bombers. Time 18:30. No clouds. (clouds drops performance a lot)

BENCH2: 87 seconds, Spitfire in Kuban summer Novorossiysk destroying four Ju-52, lot of fire and smoke. Time 18:30. No clouds.

 

The Graphics options are like the old benchmarking: High Settings, Shadows medium, Mirrors off, distant landscape x2, Horizon draw 100Km, Landscape filter Blurred, Grass Normal, Clouds Medium, DRF=1, AA x2, gamma 0.8.

 

Regarding frequencies, the Index runs at 80Hz (this is what I use) and the Reverb runs at 90Hz. I also have tried the 60Hz mode of the REverb but this is absolutely a non-go since it degrades the visual clarity a lot and also the vision is cross-eyed, that is very strange.

Running the Index at 80Hz helps not only the GPU load but also the CPU load since less scenes per second need to be generated, everything which helps CPU load (single core load) is very much welcomed in IL-2 VR. So it is bad that the Reverb only works at 90Hz, since it will demand more power from CPU and GPU.

 

Then, for the applied SS:

100%, 188% and 300% for the Reverb, this is 5, 9.5 and 15.1million pixels in total respectively at 90Hz, making 0.45, 0.85 and 1.36 billion pixels per second respectively. 

140% for the Index, this is 12.6million pixels in total at 80Hz making 1.01 billion pixels per second. (going beyond 140% might not be worth the extra detail)

 

The SteamVR recommended SS in Reverb for my PC and 90Hz is 150% (which is 7.6million pixels at 90Hz, so 0.68 billion pixels per second)

The SteamVR recommended SS in Index for my PC and 80Hz is 150% (which is 13.5 million pixels at 80Hz, so 1.08 billion pixels per second which is a much higher rate than in Reverb)

I really don´t know the algorithm used by SteamVR to make the SS recommendation, I think it uses as inputs the VR device, the GPU, the display frequency,etc,  but I really don´t know.

 

BENCH1 results: (here I also run the bench in 4K monitor at 60Hz):

In the 4K monitor we can see the overall load of the bench1. Average fps is 156.

The Index keeps almost always at 80 fps, excepts to low peaks (marine artillery hit my Spit!). Average fps is 78.2. So 97.8% close to max (80fps)

The Reverb has a little harder time to try to be at 90fps, both at 100% and 188%. I didn´t run this bench1 at 300%. Average fps is 72.3 (SS100%) and 69(SS188%). So respectively, 80.3% and 76.7% close to max (90 fps)

Here we can see that the monitor pattern is followed by the Reverb, revealing when the scenes are more demanding in terms of CPU (since my 4K at 60Hz is not constrained by GPU).

We can see also that the difference in performance from SS100% to 188% is very small, just 3.6% reduction. But the visual gain that we obtain is very large, so it is really worth to run at 188% in my system.

I also display (just below fps graph) the GPU load for the Reverb at SS188%, you will see that it is in the 50-70% band, so a 1080Ti is quite OK for the Reverb at SS=188%. (and this track).

 

bench1-Index-Reverb.thumb.png.b38b4e72741ddda412bd1d505c2d3361.png

 

BENCH2 results: (here I also run the Reverb at SS=300% just to see influence)

The Index keeps 80 almost all time except when the ablaze Ju-52 crash in ground. 

The Reverb does well at beginning but later can not keep the 90fps and slowly go down to 70s (much trailing smoke and fire).

The SS=300% is clearly a non-go in both: performance and visual gain.

 

The GPU loads of Index and REverb, show that the 12.6 million pixel of Index loads almost equally the GPU than the 9.5 million pixel of REverb at 188%.

(If we take 12.6 and multiply by 80 and divide by 90, it is equivalent to 11.2 million pixel at 90Hz, still 1.7 million more than 9.5).

 

bench2-Index-Reverb.thumb.png.e1d96f965401509781a86386f47058d9.png

 

Conclusion regarding performance:

 

Both tracks reveal something really interesting. Even if the Index in handling more pixels than REverb (both 100% and 188%) it gets better performance and keeps the fps at the desired rate of the display frequency (80Hz). Running the Index at 80Hz or 90Hz I don´t notice the difference in comfort.

Taking into account that these tracks are relatively simple (no clouds at all, medium shadows, no mirrors, just High settings), the Reverb at 188% has serious difficulty to keep 90fps, but the guilty guy is not GPU. We should look into the CPU, the 90Hz and the WMR software.

The reason for the Reverb worse performance would be the always bottleneck of CPU (trying to keep 90 frames per second) and also all the WMR software. Yes, constructing and rendering the scene 90 times per second is more demanding than doing it "only" 80 times per second. But this would not justify all the difference in performance.

The performance of the Reverb is not a question of the SS we apply, since even at SS100% the performance is bad. SS=100% in REverb it is just only 5 million pixels!!!, this is like running the CV1 Rfit at SS=118%!!!. For sure my old Rift at SS=118% and my PC would deliver solid 90fps for this bench1 track, but I already sold it, so I can not test it, but with more complex benchs from the past my Rift was in the 85-90 range.

So unless I am missing something, with IL-2, the WMR software and interface (WMR for SteamVR) is not working as optimally as the Index software (just SteamVR)

 

 

On 9/6/2019 at 2:13 AM, whitav8 said:

@ChilliWIlli,

      This is a plot of SteamVR reprojection (auto, not motionvector) on/off with my 9700K@5Ghz+RTX2070+Odyssey+   Thanks again for your detailed analysis! Seems like SteamVR reprojection should be more responsive to returning to 90Hz.

 

Thanks for your graph, it is interesting to see how "ASW like" tools works.

 

I am more interested in knowing how you have run the bench1 in terms of in-game settings and also SteamVR SS. Could you detail it please?

  • Like 2
  • Thanks 2

Share this post


Link to post
Share on other sites

@ChilliWilli,

       Great analysis and I imagine you are correct about performance. I feel that few of the developers are really carefully watching their performance with benchmarks like you are. Interesting though that there were some forum reports who have found better performance with their Reverb (than ??? - maybe DCS - and versus CV-1 with high SS??) on a different application than IL-2.

 

I use your same settings except for gamma=0.9 on the benchmarks.

 

 

Share this post


Link to post
Share on other sites
11 hours ago, whitav8 said:

I use your same settings except for gamma=0.9 on the benchmarks.

 

I am intringued by your good performance with the bench1, you were mostly at 90fps.

 

Do you use "WMR for SteamVR" beta?

What SS did you applied when running the bench1?

What is you mobo and RAM speed?

Share this post


Link to post
Share on other sites

@Chilliwilli,

       

Do you use "WMR for SteamVR" beta? - Yes

What SS did you applied when running the bench1?  I am using 156% which is about 2200x2500 (Don't forget this is just an Odyssey+ )

What is your mobo and RAM speed? ASUS ROG Maximus XI Hero with 32gb at 3200

 

9700K@5Ghz CPU and RTX2070 with mild overclock (GPU=+200Mhz  and Mem Clock = +800 Mhz   - GPU clock runs at about 2000 Mhz under heavy load with Mem clock @7800Mhz )

 

Dave W.

Share this post


Link to post
Share on other sites

Chili - here are my results with Reverb at 192%SS for the the two tracks.   Used your setting with a 9600k@5.0Ghz, DDR4@3200Mhz, and 1080@1900Mhz.

 

On test 1 looks like your theory on the CPU is right, even with only 1080 but with a faster clock speed and RAM, I was able to match or beat your FPS.

 

On test 2 I was at least 5-10 FPS below yours, either I didn't run the test right or the smoke/fire stressed my 1080 more than your 1080Ti 

Slide1.JPG

Presentation1.jpg

Share this post


Link to post
Share on other sites
3 hours ago, whitav8 said:

Do you use "WMR for SteamVR" beta? - Yes

What SS did you applied when running the bench1?  I am using 156% which is about 2200x2500 (Don't forget this is just an Odyssey+ )

What is your mobo and RAM speed? ASUS ROG Maximus XI Hero with 32gb at 3200

 

I have installed the "WMR for SteamVR" beta and run the bench1 twice and there is not measurable gain in fps (not bigger than noise measurement).

 

Certainly your CPU and RAM is quite capable and your mild overcloked 2070 do a great job in running that SS ratio,, which is bigger than the native Reverb. 

 

I don´t know if my worse performance is due to my PC (just 4.8 GHz and slower RAM) or if the WMR REverb software is less optimum than the WMR Odyssey software. If I would keep the Reverb (probably I will not) I will most likely upgrade my PC to something similar to you.

Share this post


Link to post
Share on other sites

@ChilliWilli,

       I enjoyed doing the benchmarks without ASW / Reprojection (WMR terminology ) in order to generally show performance potential. You change  one parameter like shadow quality or SuperSample value and retest to see the effect. However, most of us will be actually running a lot of the time with ASW / Reprojection enabled since for most runs, our actual  performance will be from 70 to 90 fps since we look around to see explosions, out the side windows, etc.. and we will get a lot of frame jitter if the framerate is not right at 90 or 80 (Index ). It would be nice if Reverb and maybe all HMDs could support 75fps (60 is too low ) to give a little more headroom for use without ASW / Reprojection. The quality and time efficiency of ASW / Repro is very difficult to measure - we would have to get into using FRAPS for frametimes every frame (a lot of data ) looking for deltas between frames to try to measure jitter.

 

Quoting you:  "I don´t know if my worse performance is due to my PC (just 4.8 GHz and slower RAM) or if the WMR Reverb software is less optimum than the WMR Odyssey software" - they really have to be the same (Reverb and Odyssey + ) - it's just WMR portal and SteamVR together I agree that Index might be somewhat better in performance since it is one less layer. Your system is only slightly slower than mine (4.8 Ghz vs 5 Ghz) - from what I have heard, the biggest improvement would be to have a 2080Ti. Small changes may not really do much for jitter.

 

One thing that isn't being discussed is the idea of the "Single Pass" stereo mode that Nvidia has - possibly something similar from AMD?? I know from benchmarking Prepar3D which has that mode feature, it speeds up VR execution by 30 to 50% - same graphics pipeline with slightly different eyepoint value for the second eye.

 

The other thing that I have noticed - but somewhat difficult to compare due to lack of certain features - is another flight sim Aerofly FS2 with a fairly complex aircraft panel and terrain scenery along with a complicated airport model (but simplified clouds and no other aircraft except those on the ground ) runs at 250 ->300fps on my PC monitor (nonVR) and therefore runs pretty solid at 90fps in VR. I mention it because I believe that the core graphics engines of IL-2, Prepar3D,  and DCS just aren't modern and streamlined and therefore really should be rewritten. I doubt that will happen.

 

Please keep testing and evaluating Reverb and Index - the best high resolution HMDs for now. I don't know whether to wait a while or not.

Edited by whitav8

Share this post


Link to post
Share on other sites
1 hour ago, 71st_AH_statuskuo said:

Chili - here are my results with Reverb at 192%SS for the the two tracks.   Used your setting with a 9600k@5.0Ghz, DDR4@3200Mhz, and 1080@1900Mhz.

 

On test 1 looks like your theory on the CPU is right, even with only 1080 but with a faster clock speed and RAM, I was able to match or beat your FPS.

 

On test 2 I was at least 5-10 FPS below yours, either I didn't run the test right or the smoke/fire stressed my 1080 more than your 1080Ti 

 

Thank you for running the track1 with the Reverb. Now I have a more clear idea about what is causing the bad performance, it is not my hardware but the WMR software for the REverb.

Your PC is very similar in terms of CPU to the whitav8 and he shows a much better performance using the O+.

 

Well, in fact, When I was originally running the bench1 I did a mistake (not stop a heavy background application that I was running, I din´t know it since it was running at background). So I stop it and then run the bench2.

Today, just to try the WMR for steamVR beta, I run the bench1 again (with no background apps) with the non-beta and then with the beta.

This is what I saw, which is a bit more aligned to your performance:

1641068729_becnh1reverbcleanrun.thumb.png.ca344e807a66a296c0be73451eacc05e.png

 

Regarding the bench2, the smoke/fire are at the same time than low altitude, so perhaps this section of the test load more the GPU, but I would not expect such a difference for a 1080 vs 1080Ti.  Or perhaps you were applying more than 188% (your test is very similar to my 300% test).

 

29 minutes ago, whitav8 said:

ASW / Reprojection (WMR terminology

 

Oculus was the first to create the ASW tech. That tech is called "Motion Smoothing" in SteamVR (so Valve tech), and it is called "Smartsmoothing" in the Pitool (Pimax tech).

They do the same, just create synthetic frames in between the real frames when real frames are generated below the frequency display.

 

I don´t know why people in general use the term Reprojection. Reprojection is the equivalent to Oculus ATW which is a different technique which is always on in Oculus.

https://uploadvr.com/reprojection-explained/ 

40 minutes ago, whitav8 said:

You change  one parameter like shadow quality or SuperSample value and retest to see the effect

 

Yes, benchmarks are great to measure influence of parameters. One day one person could try to run a benchmark using all combinations of settings and explain to the rest of us how much influence every parameter.

42 minutes ago, whitav8 said:

It would be nice if Reverb and maybe all HMDs could support 75fps (60 is too low )

 

The Index is great since it allows multiple freqs (80, 90, 120, 144) and also the Pimax5K+(64, 72, 90). This is something other VR manufacturers (including Oculus) could copy.

46 minutes ago, whitav8 said:

they really have to be the same (Reverb and Odyssey + )

 

Might not be exactly the same, although both are WMR devices, each one has they own drivers produced by Samsumg or by HP. If not, how you will explain the difference in the bench1 with you (best) and me&statuskuo(worse)?

50 minutes ago, whitav8 said:

from what I have heard, the biggest improvement would be to have a 2080Ti

 

I don´t think a 2080Ti will help too much here since we have seen that the difference on SS=100% and SS=188% is very small, so it means that it is not a GPU problem.

Anyone with a 2080Ti and a REverb could please run the bench1 just to see if we are right or wrong.

Share this post


Link to post
Share on other sites
22 minutes ago, chiliwili69 said:

 

I don´t know why people in general use the term Reprojection. Reprojection is the equivalent to Oculus ATW which is a different technique which is always on in Oculus.

 

 

I guess because that is what Microsoft refers to it in Windows Mixed Reality?

 

    "driver_Holographic_Experimental" : {
        // Motion reprojection doubles framerate through motion vector extrapolation

 

It is also what SteamVR refers to in settings where it says Reprojection On for the Reverb if the settings file has been edited to enable it

Share this post


Link to post
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now

×
×
  • Create New...